TW201910950A - Robot processing method and system based on 3d image - Google Patents

Abstract

Robot processing method and system based on 3D image are provided. The processing method includes the following steps: providing a robot 3D model data and a processing environment 3D model data; obtaining a workpiece 3D model data, and generating a processing path formed from multiple contacting points according to the workpiece 3D model data, wherein a free end of the robot moves according the processing path to complete the processing; generating a posture candidate group according to a relationship between the free end of the robot to any corresponding one of the contacting points; selecting a real moving posture from the posture candidate group; moving the free end of the robot to any corresponding one of the contacting points according to the selected real moving posture; and the free end of the moving in the processing path according to the real moving posture for completing the processing.

Description

Robot arm processing method and system based on three-dimensional image

The present invention relates to a processing method and system, and more particularly to a method and system for processing a robot arm based on a three-dimensional image.

Many of the processing steps in the foundry industry are single and repeated. At present, the manpower is gradually replaced by mechanical means, and the use of machines to perform processing helps to increase product output and reduce manpower expenditure. In addition, the control of the machining path through the program can reduce the uncertainty when using manual machining, so the use of machine processing becomes a better choice in the production process of various products.

In the current machine processing, such as the processing path of the dispensing processing equipment, it is usually designed by engineers. If the machining path is simple, such as a linear movement or simply flipping a single angle, the programming of the machining path program is relatively simple. However, complex machining paths, such as irregular arc movements or irregular graphic movements, are difficult to program the machining path.

In addition, for the processing machine, when the machining is performed, the workpiece is mainly moved to the position to be processed by controlling the robot arm. At this time, if the tolerance of the manufacturing tolerance is made, each single workpiece of the same style may be slightly different. The difference, or the deviation of the distance or angle of the workpiece relative to the robot arm caused by the mechanical arm gripping the workpiece, will affect the machining result.

The invention provides a three-dimensional image-based robot arm processing method and system capable of improving assembly yield.

The invention relates to a method for processing a robot arm based on a three-dimensional image, which uses a robot arm to perform a processing procedure on at least one workpiece in a processing environment. The processing method includes at least the following steps: providing three-dimensional model information of the robot arm and three-dimensional model information of the processing environment; obtaining three-dimensional model information of the workpiece, and generating a processing path formed by the plurality of contact points according to the three-dimensional model information of the workpiece, The free end of the robot arm moves according to the processing path to complete the machining process; according to the relationship between the free end of the robot arm corresponding to any one of the contact points, a moving posture candidate group of the robot arm is generated; and the moving movement candidate group selects an actual movement Attitude; moving the free end of the robot arm to any corresponding contact point according to the actual moving posture; according to a plurality of actual moving postures, the free end of the robot arm moves to the machining path to complete the machining process.

The invention further provides a robot arm processing system based on three-dimensional images, comprising: a mechanical arm, processing a workpiece on at least one workpiece in a processing environment; a database, storing three-dimensional model information of the workpiece, three-dimensional model information of the robot arm, processing a three-dimensional model information of the environment; and a processing module coupled between the robot arm and the database for controlling the robot arm to execute the machining program; wherein the processing module generates the machining formed by the plurality of contact points according to the three-dimensional model information of the workpiece a path, the free end of the control robot moves according to the machining path to complete the machining process; wherein the processing module generates a moving posture candidate group of the robot arm according to the relationship between the free ends of the robot arm corresponding to any one of the contact points, and Moving the gesture candidate group, selecting an actual moving posture; wherein the processing module controls the free end of the robot arm to move to any one of the contact points according to the actual moving posture; wherein the processing module controls the free end of the robot arm according to the plurality of actual moving postures Move to the machining path to complete the machining program

Based on the above, the three-dimensional image-based robot arm processing method and system provided by the present invention can perform an automatic processing program, which not only can save manpower, but also can reduce human error and improve assembly yield.

The above described features and advantages of the invention will be apparent from the following description.

The invention provides a robot arm processing method and system based on three-dimensional image, and inputs virtual world 3D space by correcting information such as processing environment and processing equipment in the real world, and corrects virtual 3D space and real world. Then, after obtaining the 3D information of the workpiece to be processed in the real world, and inputting the 3D information of the workpiece to be processed in the real world into the database and calculating it, cooperate with the processing equipment and the processing environment in the virtual 3D space. The machining path is then created, so that the robot in the real world can be processed in the real world according to the machining path calculated in the virtual 3D space.

[First Embodiment]

1A is a schematic view of a three-dimensional image-based robot arm processing system of the first embodiment, FIG. 1B is a schematic view of a processing path formed by a contact point of the first embodiment, and FIG. 2 is a three-dimensional image-based machine of the first embodiment. Flow chart of the arm processing method. Referring to FIG. 1A , FIG. 1B and FIG. 2 simultaneously, the three-dimensional image-based robot processing system 100 includes a processing module 110 , a robot arm 120 , and a database 130 . The robot arm 120 is configured to perform processing on at least one workpiece 190 in a processing environment; the data library 130 is configured to store three-dimensional model information of the workpiece 190, three-dimensional model information of the robot arm 120, and three-dimensional model information of the processing environment; and the processing module 110 It is electrically coupled between the robot arm 120 and the database 130 for controlling the robot arm 120 to perform a machining process.

The foregoing processing module 110 and the database 130 may be built in the same electronic device (for example, a host computer), or the database 130 and the processing module 110 may be two independent entities, such as the database 130. It can be a portable hard disk, and the portable hard disk can be electrically connected to the processing module 110 through the medium.

When the workpiece 190 is processed using the aforementioned processing system, at least the following steps S110 to S160 are included.

In step S110, three-dimensional model information of the robot arm 120 and three-dimensional model information of the processing environment are provided.

In step S120, the three-dimensional model information of the workpiece 190 is obtained. The obtaining the three-dimensional model information of the workpiece 190 may be that the processing module 110 obtains the three-dimensional model information of the preset workpiece 190 through the database 130, or may be through the non-contact type. The detecting device 140 detects the three-dimensional model information of the workpiece 190 generated by the contour and size of the workpiece 190. The workpiece 190 may be a housing of any device (including electronic devices and non-electronic devices) or a housing of a golf club head, etc., as long as it is an article to be processed, it may be a workpiece in the processing system or the processing method. It is not limited to the examples given in the embodiment. In the present embodiment, the workpiece 190 is described in the housing of the electronic device. The housing of the electronic device has a rectangular shape, and the surface to be processed is a flat surface. In addition, the embodiment of detecting the workpiece 190 through the non-contact detecting device 140 may be to capture an image of the workpiece 190 using a depth camera, or to scan the contour of the workpiece 190 by means of 3D laser scanning, or may be depth. The camera is mixed with a 3D laser scan. Of course, the embodiment of detecting the workpiece 190 through the non-contact detecting device 140 is not limited by the example of the embodiment, and those skilled in the art can select an appropriate method according to the requirements.

Incidentally, between step S110 and step S120, step S112 may be further included, and the processing module 110 corrects the error of the three-dimensional model information of the robot arm 120 and the three-dimensional model information of the processing environment in the real world coordinate system.

The foregoing correction means that the processing module 110 selects at least one calibration point coordinate position information in the three-dimensional model information of the robot arm 120 and the three-dimensional model information of the processing environment, and the processing module 110 makes information according to at least one calibration point coordinate position information. The free end of the robot arm 120 moves to a corresponding coordinate position on the real world coordinate system, and the processing module 110 compares at least one correction point position information with a corresponding coordinate position.

Please refer to FIG. 1C. The left side of FIG. 1C is the robot arm in the real world, and the right side of FIG. 1C is the robot arm in the virtual 3D space. In detail, at least one reference point S is selected on the real world robot arm to obtain three-dimensional information of the reference point S, and a comparison point C is obtained at a position of the virtual world robot arm corresponding to the reference point S, and the comparison is obtained. Point C's 3D model information. Then, the three-dimensional model information of the comparison point C is compared with the three-dimensional information of the reference point S to obtain the conversion coefficients of the real world and the virtual 3D space. Through this conversion factor, the 3D model information can be converted into real-world 3D information, enabling components in the virtual 3D space to match the same components of the real world. Of course, real-world three-dimensional information can also be converted into a three-dimensional information model in virtual 3D space through the conversion coefficient.

In particular, the machining posture and movement path of the robot arm in the virtual 3D space are further corrected with the robot arm in the real world. In detail, the real-world robotic arm is operated, and its free end is arbitrarily moved to four reference points S1, S2, S3, and S4 in the real world, and the four reference points S1, S2, S3, and S4 are mapped. The comparison points C1, C2, C3, and C4 are formed in the virtual 3D space, and the three-dimensional model information of the comparison points C1, C2, C3, and C4 is obtained. While moving the free end of the robot arm in the real world, record the moving posture and path of the free end. Then, according to the order in which the free end of the real world robot arm moves to the reference points S1, S2, S3, and S4, the free end of the robot arm in the virtual 3D space is moved to the comparison points C1, C2, C3, and C4, while The moving posture and path of the free end of the robot arm in the virtual 3D space are recorded, and compared with the moving posture and path of the robot arm in the real world, the error between the two is found and corrected.

After that, the result is transmitted back to the processing module 110, and the processing module 110 performs calculation and adjustment to make the processing posture and the moving path of the robot arm in the virtual 3D space and the processing posture and movement of the robot in the real world. The paths can be synchronized. The synchronization referred to here mainly means that the free ends of the robot arms in the real world and the virtual 3D space move along the same path in the same posture, and are not limited to the same thing at the same time, or may be different. The time to complete the same task. Simply put, it is possible to set the machining program in the virtual 3D space, and then specify that the processing system in the real world can be processed at a preset time.

In addition, after step S120, step S122 may be further included, and the processing module 110 detects at least one contact point feature 192 according to the three-dimensional model information of the workpiece to establish a position of the plurality of contact points 194, and generates a plurality of contact points 194. After the machining path is formed, after the robot arm 120 is driven by the processing module 110, the free end of the robot arm 120 can be moved according to the processing path, as shown in FIG. 1C. Incidentally, the contact points may be arranged at the same interval or at different intervals, depending on the requirements.

Then, referring to FIG. 1A, FIG. 1B and FIG. 1C, in step S130, the processing module 110 generates a motion posture candidate of the robot arm 120 according to the relationship between the free end of the robot arm 120 and the corresponding one of the contact points 194. group. The moving attitude candidate group refers to the robot arm posture corresponding to all possible paths of the machining program when the robot arm 120 moves from one place to another during the machining process.

In step S140, the actual moving posture is selected from the moving attitude candidate group. Specifically, the processing module 110 generates a corresponding three-dimensional attitude model candidate group of the robot arm 120 according to the moving posture candidate group, and the self-moving posture candidate according to the moving posture candidate group, the three-dimensional posture model candidate group, and the three-dimensional model information of the processing environment. The moving posture causing the robot arm 120 and the environmental space to interfere with each other is deleted, and a moving posture that produces a minimum offset amount to the axial angle of the robot arm 120 is selected from the moving posture candidate groups that do not interfere with each other.

In detail, when the processing module 110 calculates a plurality of moving paths in which the robot arm 120 moves to the processing position, it is necessary to consider the type, contour, or size of the robot arm 120, and also needs to be set in the processing environment. The status of other components. Without considering the above factors, the robot arm 120 may be interfered with by components in the processing environment due to its own type, contour and size during movement, rendering the machining impossible. Even the mechanical arm 120 and the components in the processing environment cause damage to the robot arm 120 or assembly due to impact.

Therefore, after calculating a plurality of possible movement paths, the processing module 110 further considers and compares the three-dimensional posture model candidate groups related to the shape, the contour, the size, and the relationship between the robot arm 120 and the workpiece 190. . At the same time, since the component configuration manner in the processing environment also affects the completion degree of the processing program, the processing module 110 also takes into consideration and compares the three-dimensional model information of the processing environment. After synthesizing and analyzing the aforementioned moving attitude candidate group, the three-dimensional attitude model candidate group, and the three-dimensional model information of the processing environment, the motion pose candidate group that may cause interference is deleted, and then the remaining components do not interfere with the components in the environment. In the moving attitude candidate group, selecting the moving posture in which the robot arm 120 has the least moving distance and the shaft rotation angle of the robot arm 120 is the smallest is not only beneficial to the mechanical arm 120 to complete the processing in the most labor-saving manner, but also helps to improve the completion degree of the machining program. .

In step S150, the processing module 110 moves the free end of the robot arm 120 to any one of the contact points 194 according to the actual moving posture. In detail, after selecting the moving posture in which the robot arm 120 can complete the machining in the least labor-saving manner, the moving posture is converted into the actual moving posture applied to the real-world coordinate system via the aforementioned conversion coefficient, and then the processing module 110 is processed. The drive robot arm 120 moves to the contact point 194 in accordance with the actual movement attitude.

In step S160, the processing module 110 moves the free end of the robot arm 120 to the machining path according to the plurality of actual movement postures to complete the machining process. Specifically, since the moving posture has been converted into the actual moving posture by the conversion coefficient, the robot arm 120 moves in the real world in accordance with the actual moving path indicated by the processing module 110 on the machining path.

In addition, through the aforementioned robot arm in the real world and the robotic arm in the virtual 3D space, it can be ensured that the robot in the real world will surely complete the machining process according to the actual moving posture, avoiding the selection in the virtual 3D space. The best machining path that comes out is not exactly the same as the actual moving path of the robot arm in the real world, and the machining program cannot be completed.

In particular, each of the workpieces 190 of the same type picked up by the robot arm 120 may have slight differences in appearance dimensions due to tolerances, or may result in the workpiece 190 being relative to the robot arm 120 due to the angle of placement of the workpiece 190. The angular deflection is such that the contact points established on each of the workpieces 190 are not identical, so the processing paths calculated by the processing module 110 for each of the workpieces 190 are not identical. Simply put, each single workpiece 190 will have its own unique processing path.

Through the above-described three-dimensional image-based robot arm processing method and system, an automated processing program can be performed, which not only saves labor, but also reduces human error and improves assembly yield.

[Second embodiment]

3 is a schematic view of the dispensing processing apparatus of the second embodiment, FIG. 4 is a flow chart of processing the housing of the golf club head by the dispensing processing apparatus, and FIG. 5 is a virtual 3D of the housing of the golf club head. Schematic diagram in space.

Referring to FIG. 3, FIG. 4 and FIG. 5 simultaneously, the dispensing processing apparatus 200 includes a processing module 110, a robot arm 120, a database 130, a non-contact detecting device 140, and a dispensing device 250. The robotic arm 120 is used to process the housing 300 of the golf club head within the processing environment 270 provided by the dispensing processing apparatus 200. The database 130 is used to store three-dimensional model information of the dispensing processing apparatus 200, the processing environment 270, and the housing 300 of the golf club head. The processing module 110 is electrically coupled between the robot arm 120 and the database 130 for controlling the robot arm 120 to perform a machining process. The non-contact detecting device 140 is, for example, an imaging lens 242 electrically coupled to the processing module 110 and the data library 130 for detecting the contour and size of the housing 300 to generate three-dimensional model information of the housing 300. The position of the dispensing device 250 can be set according to actual needs. In the present embodiment, the dispensing device 250 is disposed at a fixed position within the processing environment 270, and the free end of the robot arm 120 is coupled to the clamping device 212 and is clamped by the clamping device 212 and the robot arm 120 is along The processing path moves the housing 300, allowing the dispensing device 250 to dispense each contact point 300b on the housing 300. In another embodiment, not shown, the dispensing device 250 can be disposed at the free end of the robotic arm 120 to secure the housing 300 of the golf club head to use the robot arm 120 for each contact point on the machining path. Dispensing at 300b.

Moreover, the dispensing processing apparatus 200 further includes a laminating device 260 located within the processing environment 270 and disposed adjacent to the robotic arm 120 for providing pressure to conform the golf club head housing 300 to the golf ball. Another housing 400 of the club head in which at least one of the two housings 300, 400 of the golf club head has been dispensed. The sealing device 260 can be selected from a pneumatic cylinder, but is not limited thereto.

In addition, the dispensing processing apparatus 200 further includes a stocking area 200a and a loading area 200b, and the robot arm 120 is adapted to move between the stocking area 200a and the stocking area 200b, wherein the stocking area 200a is used to place the golf club to be processed The housing 300 of the head, and the loading zone 200b are used to place the two housings 300, 400 of the golf club head that have been affixed together.

When the dispensing process is performed on the housing 300 of the golf club head using the dispensing processing apparatus 200, the three-dimensional model information of the robot arm 120 and the three-dimensionality of the processing environment 270 are obtained in advance from the database 130 in step S210. News. The three-dimensional model information of the robot arm 120 includes the number of axes constituting the robot arm 120, the rotatable angle of the shaft, the moving direction and distance of the robot arm 120, and the like. The three-dimensional information of the processing environment 270 includes other possible components or devices other than the aforementioned non-contact detecting device 140 (camera lens 242) and the dispensing device 250. These other components or devices may be composed of dispensing processing. The assembled components of the apparatus 200, and possibly the dispensing processing apparatus 200, are intended to perform other processes.

In step S212, by comparing the error of the three-dimensional model information with the real world coordinate system, the conversion coefficient between the virtual 3D space and the real world coordinate system constructed by using the three-dimensional model information is found. By the conversion factor, the three-dimensional model of the robot arm 120 and the three-dimensional model information of the processing environment 270 can be correctly compared with the real-world robot arm 120 and the processing environment 270, and the three-dimensional model information of the correction robot 120 and the three-dimensional model of the processing environment 270 are achieved. Information, the error between the two in the real world coordinate system. Although the foregoing describes that the three-dimensional model of the robot arm 120 and the three-dimensional model information of the processing environment 270 can be correctly compared with the robot arm 120 and the processing environment 270 in the real world, those skilled in the art should also think that the conversion coefficient can be The real world robotic arm 120 and processing environment 270 are projected into a virtual 3D space. Simply put, by transforming coefficients, the coordinates in the real world can be matched to the model information of the virtual 3D world, and the model information in the virtual 3D space can also be applied to the coordinates of the real world.

In addition, the movement of the robot arm 120 in the real world and the movement of the robot arm in the virtual 3D world are further corrected to synchronize the movement of the real-world robot arm 120 with the robot arm in the virtual 3D world. The way of moving. The synchronization referred to here mainly refers to the movement of the free end of the robot arm in the real world and the virtual 3D space along the same path in the same posture, but it is not limited to the same thing at the same time, but also Complete the same task at different times.

In step S220, three-dimensional model information of the housing 300 of the golf club head is obtained through the imaging lens 242. In detail, the imaging lens 242 is disposed in a central region of the dispensing processing apparatus 200. Therefore, in actual operation, after the robot arm 120 first obtains the housing 300, the robot arm 120 moves to the adjacent camera lens 242 to allow the camera lens 242 to capture an image to obtain three-dimensional model information of the housing 300.

The three-dimensional model information of the housing 300 of the golf club head includes the contour shape, size of the housing 300 of the golf club head, the surface or curved surface to be machined, or other physical features. Of course, in another embodiment, the three-dimensional model information of the shell 300 of the golf club head can also be built in the database 130, and the processing module 110 can directly access the shell built in the database 130. 300 3D model information. Alternatively, the image of the captured housing 300 can be used in conjunction with the three-dimensional model image of the preset housing 300 in the database to generate the three-dimensional model information of the final housing 300.

The foregoing manner of incorporating the three-dimensional model information of the housing 300 into the database 130 assumes that the individual single housings 300 of the same type are completely identical and are not affected by tolerances. However, in the actual manufacturing process, the individual single housings 300 of the same kind are inevitably different due to tolerances. Therefore, the advantage of photographing each of the housings 300 that are about to enter the stocking area 200a through the imaging lens 242 is that The same type of single housing 300 performs instant feature recognition, which is advantageous for design optimization of the processing path.

In addition, the imaging lens 242 of the present embodiment is disposed on the path of the housing 300 of the golf club head moving from the preparation area 200a to the loading area 200b, but the imaging lens 242 can also be disposed on the dispensing processing equipment according to requirements. Appropriate place in 200 to capture the housing 300 via the robot arm 120 and move to the imaging lens 242 for shooting.

In step S222, the processing module 110 detects at least one contact point feature 300a according to the three-dimensional model information of the housing 300 to establish a position of the plurality of contact points 300b. In detail, the contact point feature 300a may be a single feature set manually, such as a certain recessed point, a protruding point or an edge reference point of the housing 300, or may be built into the data of the plurality of contact point features 300a. In the library 130, one of the hashes is then selected via the processing module 110 as a reference. Thereafter, the processing module 110 selects a plurality of locations on the surface to be processed as the other contact points 300b based on the selected contact point features 300a. In other embodiments, the user may also set a plurality of contact points 300 to form a predetermined processing path.

Incidentally, although the housing 300 of the golf club head is used as the workpiece to be processed, it should be understood by those skilled in the art that the type of the workpiece is not limited by this embodiment, and it is more likely to pass the camera lens. The image captured by the 242 is matched with the detected contact point feature 300a to determine the type of the workpiece to be processed, and then the processing module 110 finds corresponding information from the database 130 according to the identification result to perform a corresponding processing process.

Referring to FIG. 3, FIG. 4 and FIG. 5, when the mechanical arm 120 is gripped by the clamping device 212 connected to the free end thereof, the housing 300 placed in the preparation area 200a may be placed due to the manufacturing tolerance of the housing 300. The difference in the placement position or the placement angle causes the clamping device 212 to grip each of the single housings 300 in a different manner. Therefore, in step S230, the processing module 110 can generate a moving attitude candidate group of the robot arm 120 according to the relationship between the free end of the robot arm 120 and the corresponding one of the contact points 300b according to the three-dimensional model information of the robot arm 120. .

In detail, after the image is captured by the camera lens 242, the processing module 110 can calculate whether there is a possibility of a distance offset or an angular deflection between a certain end of the free end of the robot arm 120 and the corresponding contact point 300b. The offset distance and the deflection angle are further calculated and compensated by changing the posture of the robot arm 120, which includes the moving distance of the robot arm 120 with respect to a certain reference point, and the clamping of each axis of the robot arm 120 with each other. The angle, the relative rotation angle between the two axes, and the angle of the clamping device 212 with respect to the shaft, the angle of rotation, and the like. Therefore, the information of the moving attitude candidate group includes the distance compensation of the robot arm 120 with respect to a certain reference point, the angle compensation, and the machining angle when the mechanical arm 120 moves the housing 300 to the dispensing device 250, and the robot arm 120 moves from a certain point. All possible moving attitudes and machining paths to another fixed point...etc.

In step S240, the actual moving posture is selected from the moving attitude candidate group. In short, it is the best posture (actual movement posture) in which the moving distance of the robot arm 120 is the shortest and the rotation angle of each axis of the robot arm 120 is the smallest, from all possible moving postures.

Specifically, the processing module 110 generates a corresponding three-dimensional attitude model candidate group of the robot arm 120 according to the moving posture candidate group, wherein the three-dimensional attitude model candidate group includes the housing 300 calculated by the processing module 110 relative to the point. The glue device 250 performs an optimized angle at the time of dispensing and an optimized posture of the robot arm 120. In addition, the processing module 110 also comprehensively considers the three-dimensional model information of the moving attitude candidate group, the three-dimensional attitude model candidate group, and the processing environment 270, and calculates whether the moving path of the moving attitude candidate group interferes with other components in the processing environment 270, and The moving postures causing the robot arm 120 and the environmental space to interfere with each other are deleted from the moving attitude candidate group, and then the moving distance candidate group that has not interfered with each other is selected to have the shortest moving distance to the robot arm 120 and the minimum offset of the shaft angle. Move gesture.

In step S250, the processing module 110 converts the calculated mobile gesture calculated in the virtual 3D space into a real moving posture in the real world by the conversion coefficient, and the posture includes the machining movement path of the robot arm 120 and the shaft rotation angle. Wait.

Incidentally, due to the influence of the manufacturing tolerance of each of the single casings 300, the position or angle of the casing 300 in the stocking area 200a after entering the stocking area 200a causes the clip attached to the free end of the robot arm 120. The angle at which the holding device 212 grips the housing 300 may also need to be changed, so the actual movement attitude will be different for each single housing 300. In other words, looking at the processing of all the housings 300, the processing path of each of the single housings 300 may be slightly different, not identical.

In step S260, the processing module 110 sends a signal according to the actual moving posture to drive the free end of the robot arm 120 to start. After the housing 300 is clamped by the clamping device 212 connected to the free end, the free end of the robot arm 120 is moved according to the actual movement. The gesture moves over the processing path and approaches the dispensing device 250 at an optimal angle for dispensing. After the dispensing is completed, the robot arm 120 bonds the housing 300 to the other auxiliary housing 400, and then the two housings 300, 400 that are bonded together are placed in the loading area 200b. In addition, pressure can be applied to the two housing bodies 300, 400 that are bonded together via the bonding device 260 disposed in the loading area 200b, so that the two housing bodies 300, 400 are closely fitted together to form a golf ball. Head.

Incidentally, through the aforementioned robot arm 120 in the real world and the robot arm in the virtual 3D space, it can be ensured that the robot arm 120 in the real world can surely complete the machining program according to the actual moving posture, avoiding the virtual 3D. The optimal machining path selected in the space is not exactly the same as the actual movement path of the robot arm 120 in the real world, and the machining process cannot be completed.

Above, the machining program is completed. The golf club head can then be removed from the stocking zone 200b either manually or mechanically.

[Third embodiment]

This embodiment is substantially the same as the foregoing second embodiment except that it is in the third embodiment as shown in FIG. The housing 300 of the golf club head can be transported into the stocking area 200a through the conveyor belt P, and the camera lens 242 can be disposed in the transport path of the conveyor belt P and positioned in front of the stock preparation area 200a to place the shell of the golf club head The body 300 is photographed before entering the stocking area 200a, and then the housing 300 of the golf club head is processed into three-dimensional model information by image processing.

The remaining steps of obtaining the three-dimensional model information of the casing 300 to be processed are the same as those of the foregoing embodiment, and therefore will not be described again.

Although the present embodiment and the second embodiment are slightly different in the structure of the system and the application steps of the process, the present invention does not exceed the architecture of the three-dimensional image-based robot arm processing method and system of the present invention.

In summary, in the method and system for processing a three-dimensional image based robot of the present invention, the model is built in a virtual 3D space, and the virtual 3D space and the real world are connected by correction, in addition to being able to The processing module performs calculation and screening of the optimized processing path, and the robot arm also has the functions of automatic judgment and learning. In addition, no human intervention is required during the process, which saves manpower. Furthermore, through the automatic learning function, it is possible to correct the deviation angle or distance immediately, which is beneficial to improve the assembly yield of the workpiece.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

100‧‧‧3D image based robotic arm processing system

110‧‧‧Processing module

120‧‧‧ Robotic arm

130‧‧‧Database

140‧‧‧ Non-contact detection device

190‧‧‧Workpiece

200‧‧‧ Dispensing equipment

212‧‧‧Clamping device

242‧‧‧ camera lens

250‧‧‧ Dispensing device

260‧‧‧Fitting device

270‧‧‧Processing environment

300, 400‧‧‧ shell

200a‧‧‧Material area

200b‧‧‧Matching area

192, 300a‧‧‧ contact point characteristics

194, 300b‧‧‧ touch points

P‧‧‧ conveyor belt

S‧‧‧ benchmark

C‧‧‧ comparison point

S110~S160, S210~S260‧‧‧ steps

1A is a schematic view of a three-dimensional image-based robotic arm processing system of the first embodiment. Fig. 1B is a schematic view showing a processing path formed by the contact points of the first embodiment. FIG. 1C is a schematic diagram of coordinate conversion of a virtual 3D space and a real world coordinate system. 2 is a flow chart of a three-dimensional image-based robot arm processing method of the first embodiment. Fig. 3 is a schematic view of the dispensing processing apparatus of the second embodiment. 4 is a flow chart of a method of processing a dispensing processing apparatus. Figure 5 is a schematic illustration of the housing of the golf club head in a virtual 3D space. Figure 6 is a schematic illustration of another embodiment of a dispensing processing apparatus.

Claims (23)

A machining method for a robot arm based on a three-dimensional image, using a robot arm to perform a machining process on at least one workpiece in a processing environment, the processing method comprising: providing a three-dimensional model information of the robot arm and a three-dimensional model of the machining environment Obtaining the three-dimensional model information of the workpiece, and generating a processing path formed by the plurality of contact points according to the three-dimensional model information of the workpiece, wherein the free end of the robot arm moves according to the processing path to complete the processing a program for generating a motion posture candidate group of the robot arm according to a relationship between the free end of the robot arm and any one of the contact points; selecting an actual movement posture from the motion posture candidate group; according to the actual movement posture, Moving the free end of the robot arm to any one of the contact points; according to the plurality of actual movement postures, the free end of the robot arm moves to the machining path to complete the machining process. The robot arm processing method according to claim 1, further comprising: correcting the three-dimensional model information of the robot arm and the three-dimensional model information of the processing environment, and the error of the two in the real world coordinate system. The method of processing a robot arm according to claim 2, wherein the correcting step comprises: selecting at least one calibration point coordinate position information in the three-dimensional model information of the robot arm and the three-dimensional model information of the processing environment; A correction point coordinate position information, the free end of the robot arm is moved to a corresponding coordinate position on the real world coordinate system; and the at least one correction point position information and the corresponding coordinate position are compared. The method for processing a robot arm according to claim 1, wherein the obtaining the three-dimensional model information of the workpiece comprises: obtaining a preset three-dimensional model information of the workpiece through a database, or transmitting through a non-contact detection The device detects the contour and size of the workpiece and generates three-dimensional model information of the workpiece. The processing method of the robot arm according to claim 1, wherein the generating the machining path formed by the plurality of contact points comprises: setting a position of the plurality of contact points according to the three-dimensional model information of the workpiece to form the Processing path. The processing method of the robot arm according to claim 1, wherein the generating the processing path formed by the plurality of contact points comprises: detecting at least one contact point feature according to the three-dimensional model information of the workpiece to establish a plurality of The location of the contact point. The robot arm processing method according to claim 1, wherein the selecting the actual moving posture from the moving posture candidate group comprises: generating a corresponding three-dimensional posture model candidate group of the robot arm according to the moving posture candidate group And removing, according to the moving attitude candidate group, the three-dimensional attitude model candidate group, and the three-dimensional model information of the processing environment, the moving posture that causes the robot arm to interfere with the environmental space from the moving attitude candidate group. The robot arm processing method according to claim 7, wherein the selecting the actual movement posture from the moving posture candidate group comprises: selecting an axis angle of the robot arm from the movement posture candidate group that does not interfere with each other The movement pose that produces the smallest offset. The robot arm processing method according to claim 1, wherein the machining program comprises a one-step rubber processing program, wherein a free end of the robot arm is connected to a glue device, so that the robot arm is on each of the processing paths. The contact point is dispensed. The robot arm processing method according to claim 1, wherein the machining program comprises a one-step rubber processing program, wherein a free end of the robot arm is coupled to a clamping device, and the holding device is used to clamp and move the workpiece, so that The robot arm dispenses each of the contact points on the processing path through a dispensing device in a fixed position. The method for processing a robot arm according to claim 9 or 10, further comprising: attaching the glued workpiece to an auxiliary workpiece through a bonding device. A robotic arm processing system based on a three-dimensional image, comprising: a robot arm for performing a machining process on at least one workpiece in a processing environment; a database for storing three-dimensional model information of the workpiece, and three-dimensional model information of the robot arm And a processing module coupled between the robot arm and the database for controlling the robot to execute the machining program; wherein the processing module is based on the three-dimensional model information of the workpiece And generating a processing path formed by the plurality of contact points, controlling the free end of the robot arm to move according to the processing path to complete the processing procedure; wherein the processing module is based on the free end of the robot arm and the corresponding a relationship between the contact points, generating a moving attitude candidate group of the robot arm, and selecting an actual moving posture from the moving attitude candidate group; wherein the processing module controls the robot arm according to the actual moving posture The free end moves to any one of the corresponding contact points; wherein the processing module is based on a plurality of The actual moving attitude controls the free end of the robot arm to move to the machining path to complete the machining process. The robotic arm processing system of claim 12, wherein the processing module corrects the three-dimensional model information of the robot arm and the three-dimensional model information of the processing environment, and the error between the two in the real world coordinate system. The robot arm processing system of claim 13, wherein the processing module selects at least one calibration point coordinate position information in the three-dimensional model information of the robot arm and the three-dimensional model information of the processing environment; wherein the processing The module moves the free end of the robot arm to a corresponding coordinate position on the real world coordinate system according to the at least one calibration point coordinate position information; wherein the processing module compares the at least one correction point position information with the corresponding coordinate position . The robot arm processing system of claim 12, wherein the processing module passes through the database to obtain a preset three-dimensional model information of the workpiece. The robot arm processing system of claim 12, further comprising a non-contact detecting device coupled between the processing module and the database to detect the contour and size of the workpiece to generate a three-dimensional model of the workpiece News. The robot arm processing system of claim 12, wherein the processing module sets a plurality of locations of the contact points according to the three-dimensional model information of the workpiece to form the processing path. The robot arm processing system of claim 12, wherein the processing module detects at least one contact point feature based on the three-dimensional model information of the workpiece to establish a plurality of locations of the contact points. The robot arm processing system of claim 12, wherein the processing module generates a corresponding three-dimensional attitude model candidate group of the robot arm according to the moving posture candidate group; wherein the processing module is based on the moving posture candidate The group, the three-dimensional attitude model candidate group, and the three-dimensional model information of the processing environment, from the moving attitude candidate group, deleting the moving posture that causes the robot arm to interfere with the environment space. The robotic arm machining system of claim 19, wherein the processing module selects the movement attitude that produces a minimum offset from the axial angle of the robot arm from the set of movement attitude candidates that do not interfere with each other. The robotic arm processing system of claim 12, wherein the free end of the robot arm is coupled to a dispensing device such that the robot arm dispenses each of the contact points on the processing path. The robotic arm processing system of claim 12, wherein the free end of the robot arm is coupled to a clamping device, and the clamping device is used to clamp and move the workpiece to dispense the mechanical arm through one of the fixed positions. The device dispenses each of the contact points on the processing path. The robotic arm processing system of claim 21 or 22, further comprising: a fitting device located in the processing environment and adjacent to the robot arm to attach the glued workpiece to an auxiliary workpiece.